Fix reverse CMap lookup for ToUnicode map generation

git-svn-id: svn://tug.org/texlive/trunk@34744 c570f23f-e606-0410-a88d-b1316a301751

7 files changed, 131 insertions, 29 deletions

diff --git a/Build/source/texk/dvipdfm-x/ChangeLog b/Build/source/texk/dvipdfm-x/ChangeLog
index b80bd705cc4..14e55b48ace 100644
--- a/Build/source/texk/dvipdfm-x/ChangeLog
+++ b/Build/source/texk/dvipdfm-x/ChangeLog
@@ -3,6 +3,9 @@
 	* cff.c, cff.h, cid.c, cid_p.h, cidtype0.c, fontmap.c, fontmap.h,
 	pdfdev.c: Fix CID-keyed fonts glyph lookup with XeTeX.
 
+	* pdfdev.c, pdffont.c, pdffont.h, tt_cmap.c, type0.c, type0.h: Fix
+	reverse CMap lookup for ToUnicode map generation.
+
 2014-07-26  Peter Breitenlohner  <peb@mppmu.mpg.de>
 
 	* dpxfile.c: Add a default case to avoid compiler warnings.
diff --git a/Build/source/texk/dvipdfm-x/pdfdev.c b/Build/source/texk/dvipdfm-x/pdfdev.c
index 6fa42237542..b7d2d9a5b82 100644
--- a/Build/source/texk/dvipdfm-x/pdfdev.c
+++ b/Build/source/texk/dvipdfm-x/pdfdev.c
@@ -463,6 +463,7 @@ struct dev_font {
 
   pdf_obj *resource;
   char    *used_chars;
+  char    *used_glyphs;
 
   /* Font format:
    * simple, composite or bitmap.
@@ -876,6 +877,8 @@ dev_set_font (int font_id)
   if (!real_font->resource) {
     real_font->resource   = pdf_get_font_reference(real_font->font_id);
     real_font->used_chars = pdf_get_font_usedchars(real_font->font_id);
+    if (real_font->cff_charsets)
+      real_font->used_glyphs = pdf_get_font_usedglyphs(real_font->font_id);
   }
 
   if (!real_font->used_on_this_page) { 
@@ -1155,6 +1158,11 @@ pdf_dev_set_string (spt_t xpos, spt_t ypos,
   length  = instr_len;
 
   if (font->format == PDF_FONTTYPE_COMPOSITE) {
+    if (real_font->used_glyphs != NULL) {
+      for (i = 0; i < length; i += 2)
+        add_to_used_chars2(real_font->used_glyphs,
+                           (unsigned short) (str_ptr[i] << 8)|str_ptr[i+1]);
+    }
     if (handle_multibyte_string(font, &str_ptr, &length, ctype) < 0) {
       ERROR("Error in converting input string...");
       return;
@@ -1536,6 +1544,7 @@ pdf_dev_locate_font (const char *font_name, spt_t ptsize)
 
   font->resource   = NULL; /* Don't ref obj until font is actually used. */  
   font->used_chars = NULL;
+  font->used_glyphs = NULL;
 
   font->extend     = 1.0;
   font->slant      = 0.0;
diff --git a/Build/source/texk/dvipdfm-x/pdffont.c b/Build/source/texk/dvipdfm-x/pdffont.c
index e25a1ccc9e9..f0189d74195 100644
--- a/Build/source/texk/dvipdfm-x/pdffont.c
+++ b/Build/source/texk/dvipdfm-x/pdffont.c
@@ -334,6 +334,24 @@ pdf_get_font_usedchars (int font_id)
   }
 }
 
+char *
+pdf_get_font_usedglyphs (int font_id)
+{
+  pdf_font *font;
+
+  CHECK_ID(font_id);
+
+  font = GET_FONT(font_id);
+  if (font->subtype == PDF_FONT_FONTTYPE_TYPE0) {
+    Type0Font *t0font;
+
+    t0font = Type0Font_cache_get(font->font_id);
+    return Type0Font_get_usedglyphs(t0font);
+  }
+
+  return NULL;
+}
+
 int
 pdf_get_font_wmode (int font_id)
 {
diff --git a/Build/source/texk/dvipdfm-x/pdffont.h b/Build/source/texk/dvipdfm-x/pdffont.h
index a833fe89dc1..38ca3a590e2 100644
--- a/Build/source/texk/dvipdfm-x/pdffont.h
+++ b/Build/source/texk/dvipdfm-x/pdffont.h
@@ -63,6 +63,7 @@ extern int      pdf_font_findresource  (const char *font_name,
 extern int      pdf_get_font_subtype   (int font_id);
 extern pdf_obj *pdf_get_font_reference (int font_id);
 extern char    *pdf_get_font_usedchars (int font_id);
+extern char    *pdf_get_font_usedglyphs (int font_id);
 
 #if 0
 extern char    *pdf_get_font_fontname  (int font_id); /* without unique tag */
diff --git a/Build/source/texk/dvipdfm-x/tt_cmap.c b/Build/source/texk/dvipdfm-x/tt_cmap.c
index 718e8586ae6..f49421e910a 100644
--- a/Build/source/texk/dvipdfm-x/tt_cmap.c
+++ b/Build/source/texk/dvipdfm-x/tt_cmap.c
@@ -941,6 +941,31 @@ handle_subst_glyphs (CMap *cmap,
   return count;
 }
 
+static cff_font *
+prepare_CIDFont_from_sfnt(sfnt* sfont)
+{
+  cff_font *cffont;
+  unsigned long offset = 0;
+
+  if (sfont->type != SFNT_TYPE_POSTSCRIPT     ||
+      sfnt_read_table_directory(sfont, 0) < 0 ||
+      (offset = sfnt_find_table_pos(sfont, "CFF ")) == 0) {
+    return NULL;
+  }
+
+  cffont = cff_open(sfont->stream, offset, 0);
+  if (!cffont)
+    return NULL;
+
+  if (!(cffont->flag & FONTTYPE_CIDFONT)) {
+    cff_close(cffont);
+    return NULL;
+  }
+
+  cff_read_charsets(cffont);
+  return cffont;
+}
+
 static pdf_obj *
 create_ToUnicode_cmap4 (struct cmap4 *map,
 			const char *cmap_name, CMap *cmap_add,
@@ -952,6 +977,7 @@ create_ToUnicode_cmap4 (struct cmap4 *map,
   USHORT    c0, c1, gid, count, ch;
   USHORT    i, j, d, segCount;
   char      used_glyphs_copy[8192];
+  cff_font *cffont = prepare_CIDFont_from_sfnt(sfont);
 
   cmap = CMap_new();
   CMap_set_name (cmap, cmap_name);
@@ -979,10 +1005,11 @@ create_ToUnicode_cmap4 (struct cmap4 *map,
 	       map->idDelta[i]) & 0xffff;
       }
       if (is_used_char2(used_glyphs_copy, gid)) {
+        unsigned int cid = cffont ? cff_charsets_lookup_inverse(cffont, gid) : gid;
 	count++;
 
-	wbuf[0] = (gid >> 8) & 0xff;
-	wbuf[1] = (gid & 0xff);
+	wbuf[0] = (cid >> 8) & 0xff;
+	wbuf[1] = (cid & 0xff);
 
 	wbuf[2] = (ch >> 8) & 0xff;
 	wbuf[3] =  ch & 0xff;
@@ -1014,6 +1041,9 @@ create_ToUnicode_cmap4 (struct cmap4 *map,
   }
   CMap_release(cmap);
 
+  if (cffont)
+    cff_close(cffont);
+
   return stream;
 }
 
@@ -1029,6 +1059,7 @@ create_ToUnicode_cmap12 (struct cmap12 *map,
   ULONG     i, ch;
   USHORT    gid, count;
   char      used_glyphs_copy[8192];
+  cff_font *cffont = prepare_CIDFont_from_sfnt(sfont);
 
   cmap = CMap_new();
   CMap_set_name (cmap, cmap_name);
@@ -1049,9 +1080,10 @@ create_ToUnicode_cmap12 (struct cmap12 *map,
       d   = ch - map->groups[i].startCharCode;
       gid = (USHORT) ((map->groups[i].startGlyphID + d) & 0xffff);
       if (is_used_char2(used_glyphs_copy, gid)) {
+        unsigned int cid = cffont ? cff_charsets_lookup_inverse(cffont, gid) : gid;
         count++;
-        wbuf[0] = (gid >> 8) & 0xff;
-        wbuf[1] = (gid & 0xff);
+        wbuf[0] = (cid >> 8) & 0xff;
+        wbuf[1] = (cid & 0xff);
         len = UC_sput_UTF16BE((long)ch, &p, wbuf+WBUF_SIZE);
 
         CMap_add_bfchar(cmap, wbuf, 2, wbuf+2, len);
@@ -1081,6 +1113,9 @@ create_ToUnicode_cmap12 (struct cmap12 *map,
   }
   CMap_release(cmap);
 
+  if (cffont)
+    cff_close(cffont);
+
   return stream;
 }
 
diff --git a/Build/source/texk/dvipdfm-x/type0.c b/Build/source/texk/dvipdfm-x/type0.c
index ec195772c08..e1f1feeb954 100644
--- a/Build/source/texk/dvipdfm-x/type0.c
+++ b/Build/source/texk/dvipdfm-x/type0.c
@@ -85,6 +85,7 @@ struct Type0Font {
   char    *fontname;   /* BaseFont */
   char    *encoding;   /* "Identity-H" or "Identity-V" (not ID) */
   char    *used_chars; /* Used chars (CIDs) */
+  char    *used_glyphs; /* Used glyphs (GIDs) */
 
   /*
    * Type0 only
@@ -112,6 +113,7 @@ Type0Font_init_font_struct (Type0Font *font)
   font->descriptor = NULL;
   font->encoding   = NULL;
   font->used_chars = NULL;
+  font->used_glyphs = NULL;
   font->descendant = NULL;
   font->wmode      = -1;
   font->flags      = FLAG_NONE;
@@ -131,6 +133,8 @@ Type0Font_clean (Type0Font *font)
       ERROR("%s: FontDescriptor unexpected for Type0 font.", TYPE0FONT_DEBUG_STR);
     if (!(font->flags & FLAG_USED_CHARS_SHARED) && font->used_chars)
       RELEASE(font->used_chars);
+    if (font->used_glyphs)
+      RELEASE(font->used_glyphs);
     if (font->encoding)
       RELEASE(font->encoding);
     if (font->fontname)
@@ -139,6 +143,7 @@ Type0Font_clean (Type0Font *font)
     font->indirect   = NULL;
     font->descriptor = NULL;
     font->used_chars = NULL;
+    font->used_glyphs = NULL;
     font->encoding   = NULL;
     font->fontname   = NULL;
   }
@@ -147,13 +152,45 @@ Type0Font_clean (Type0Font *font)
 /* PLEASE FIX THIS */
 #include "tt_cmap.h"
 
+pdf_obj *Type0Font_create_ToUnicode_stream(Type0Font *font) {
+  CIDFont *cidfont = font->descendant;
+  char *used = Type0Font_get_usedglyphs(font);
+  if (!used)
+    used = Type0Font_get_usedchars(font);
+
+  return otf_create_ToUnicode_stream(CIDFont_get_ident(cidfont),
+                                     CIDFont_get_opt_index(cidfont),
+                                     used);
+}
+
+/* Try to load ToUnicode CMap from file system first, if not found fallback to
+ * font CMap reverse lookup. */
+pdf_obj *Type0Font_try_load_ToUnicode_stream(Type0Font *font, char *cmap_base) {
+  char *cmap_name = NEW(strlen(cmap_base) + strlen("-UTF-16"), char);
+  pdf_obj *tounicode;
+
+  sprintf(cmap_name, "%s-UTF16", cmap_base);
+  tounicode = pdf_read_ToUnicode_file(cmap_name);
+  if (!tounicode) {
+    sprintf(cmap_name, "%s-UCS2", cmap_base);
+    tounicode = pdf_read_ToUnicode_file(cmap_name);
+  }
+
+  RELEASE(cmap_name);
+
+  if (!tounicode)
+    tounicode = Type0Font_create_ToUnicode_stream(font);
+
+  return tounicode;
+}
+
 static void
 add_ToUnicode (Type0Font *font)
 {
   pdf_obj    *tounicode;
   CIDFont    *cidfont;
   CIDSysInfo *csi;
-  char       *cmap_name, *fontname;
+  char       *fontname;
 
   /*
    * ToUnicode CMap:
@@ -197,39 +234,24 @@ add_ToUnicode (Type0Font *font)
     switch (CIDFont_get_subtype(cidfont)) {
     case CIDFONT_TYPE2:
       /* PLEASE FIX THIS */
-      tounicode = otf_create_ToUnicode_stream(CIDFont_get_ident(cidfont),
-					      CIDFont_get_opt_index(cidfont),
-					      Type0Font_get_usedchars(font));
+      tounicode = Type0Font_create_ToUnicode_stream(font);
       break;
     default:
       if (CIDFont_get_flag(cidfont, CIDFONT_FLAG_TYPE1C)) { /* FIXME */
-	tounicode = otf_create_ToUnicode_stream(CIDFont_get_ident(cidfont),
-						CIDFont_get_opt_index(cidfont),
-						Type0Font_get_usedchars(font));
+        tounicode = Type0Font_create_ToUnicode_stream(font);
       } else if (CIDFont_get_flag(cidfont, CIDFONT_FLAG_TYPE1)) { /* FIXME */
 	/* Font loader will create ToUnicode and set. */
 	return;
       } else {
-	cmap_name = NEW(strlen(fontname) + 7, char);
-	sprintf(cmap_name, "%s-UTF16", fontname);
-	tounicode = pdf_read_ToUnicode_file(cmap_name);
-	if (!tounicode) {
-	  sprintf(cmap_name, "%s-UCS2", fontname);
-	  tounicode = pdf_read_ToUnicode_file(cmap_name);
-	}
-	RELEASE(cmap_name);
+        tounicode = Type0Font_try_load_ToUnicode_stream(font, fontname);
       }
       break;
     }
   } else {
-    cmap_name = NEW(strlen(csi->registry)+strlen(csi->ordering)+8, char);
-    sprintf(cmap_name, "%s-%s-UTF16", csi->registry, csi->ordering);
-    tounicode = pdf_read_ToUnicode_file(cmap_name);
-    if (!tounicode) {
-      sprintf(cmap_name, "%s-%s-UCS2", csi->registry, csi->ordering);
-      tounicode = pdf_read_ToUnicode_file(cmap_name);
-    }
-    RELEASE(cmap_name);
+    char *cmap_base = NEW(strlen(csi->registry) + strlen(csi->ordering) + 2, char);
+    sprintf(cmap_base, "%s-%s", csi->registry, csi->ordering);
+    tounicode = Type0Font_try_load_ToUnicode_stream(font, cmap_base);
+    RELEASE(cmap_base);
   }
 
   if (tounicode) {
@@ -304,6 +326,14 @@ Type0Font_get_usedchars (Type0Font *font)
   return font->used_chars;
 }
 
+char *
+Type0Font_get_usedglyphs (Type0Font *font)
+{
+  ASSERT(font);
+
+  return font->used_glyphs;
+}
+
 pdf_obj *
 Type0Font_get_resource (Type0Font *font)
 {
@@ -481,6 +511,7 @@ Type0Font_cache_find (const char *map_name, int cmap_id, fontmap_opt *fmap_opt)
    */
 
   font->used_chars = NULL;
+  font->used_glyphs = NULL;
   font->flags      = FLAG_NONE;
 
   switch (CIDFont_get_subtype(cidfont)) {
@@ -492,11 +523,14 @@ Type0Font_cache_find (const char *map_name, int cmap_id, fontmap_opt *fmap_opt)
     /*
      * Need used_chars to write W, W2.
      */
-    if ((parent_id = CIDFont_get_parent_id(cidfont, wmode ? 0 : 1)) < 0)
+    if ((parent_id = CIDFont_get_parent_id(cidfont, wmode ? 0 : 1)) < 0) {
       font->used_chars = new_used_chars2();
-    else {
+      if (fmap_opt->cff_charsets)
+        font->used_glyphs = new_used_chars2();
+    } else {
       /* Don't allocate new one. */
       font->used_chars = Type0Font_get_usedchars(Type0Font_cache_get(parent_id));
+      font->used_glyphs = Type0Font_get_usedglyphs(Type0Font_cache_get(parent_id));
       font->flags     |= FLAG_USED_CHARS_SHARED;
     }
     break;
diff --git a/Build/source/texk/dvipdfm-x/type0.h b/Build/source/texk/dvipdfm-x/type0.h
index 817f65d9afd..da75a2ee87b 100644
--- a/Build/source/texk/dvipdfm-x/type0.h
+++ b/Build/source/texk/dvipdfm-x/type0.h
@@ -36,6 +36,8 @@ extern char      *Type0Font_get_encoding  (Type0Font *font);
 #endif
 extern char      *Type0Font_get_usedchars (Type0Font *font);
 
+extern char      *Type0Font_get_usedglyphs (Type0Font *font);
+
 extern pdf_obj   *Type0Font_get_resource  (Type0Font *font);
 
 extern void       Type0Font_set_ToUnicode (Type0Font *font, pdf_obj *cmap_ref);